Embedded Youtube
T
Timothy Coxon
To clarify, This is not a duplicate of the ability to embed a youtube video into a notebook.
When you insert a youtube video into Recall.it it appears to grab the transcript from the youtube video with time stamps and then creates links to that video with the timecodes included in the url for quicker access to that place in the video.
This is great, But I think it could be better. Rather than have the links open in a new tab as it currently does would it be possible to have it open in page as a modal popup with the youtube video embedded? Or possible a better option have a "preview tab" on the right pane so that you could have the transcript open in the left pane and be able to scan through the transcript on the left and skip directly to the appropriate point on the right hand pane.
Bonus feature!
By having the youtube video embedded like this with control over the embedded video it might actually be possible to screengrab frames from the youtube video to insert into the notebook of the note.
(( I just did a little research into this. CORS (as always) is a pain in the butt... Google locks this down directly which would leave either a serverside python script or you could add something to the chrome plugin which the page could talk to and have the chrome plugin take the screenshot itself.))
Bonus Bonus feature!
Add OCR for the screen grabs.
The ingestion of youtube videos seem to currently be based purely on the transcript download and ignores anything in the video beyond that text. With the above it would give us the ability to pull images from the videos... charts, infographics. etc.
Log In
s
schroell
I totally get this use case because I use Recall a lot for YouTube videos. This would genuinely be a nice addition.
I'd also want to add that it should be possible to redo the transcription of important YouTube videos in Recall using a better model — because the transcription YouTube provides is so cheap, shallow, and full of errors that it obviously has a huge impact on the quality of the information stored with that video. And it indirectly affects everything else too:
- all research results
- all searches
- the MCP interface
- everything
That's one of the biggest shortcomings I see with this whole YouTube transcription topic — it's done with a very fast and very cheap model on YouTube's end, and Recall just blindly relies on that. From my perspective, that was understandable at the beginning of Recall due to model costs. But I think that at least now, in the Max subscription — which really isn't cheap — there should be the option as an additional feature to redo the transcription of YouTube videos with a better model. That would bring the reader to a completely new quality level, improve the summaries, and ultimately improve the research results as well.